US20040234150A1 - Quantization matrix adjusting method for avoiding underflow of data - Google Patents
Quantization matrix adjusting method for avoiding underflow of data Download PDFInfo
- Publication number
- US20040234150A1 US20040234150A1 US10/442,280 US44228003A US2004234150A1 US 20040234150 A1 US20040234150 A1 US 20040234150A1 US 44228003 A US44228003 A US 44228003A US 2004234150 A1 US2004234150 A1 US 2004234150A1
- Authority
- US
- United States
- Prior art keywords
- quantization matrix
- adjusting method
- quantizer scale
- mode
- estq
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/136—Incoming video signal characteristics or properties
- H04N19/14—Coding unit complexity, e.g. amount of activity or edge presence estimation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
- H04N19/126—Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/172—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a picture, frame or field
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/176—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a block, e.g. a macroblock
Definitions
- the invention relates to a quantization matrix adjusting method, and more particularly to a quantization matrix adjusting method for avoiding underflow of video buffer verifier (VBV) by scaling up the quantization matrix.
- VBV video buffer verifier
- Compressing methods are typically used to compress the pictures in order to reduce the data amounts, for example coding pictures by MPEG.
- the basic unit of coding within a picture is the macroblock. If the sampling is in 4:2:0 format, there are six blocks within a macroblock: four Y blocks, one Cb block and one Cr block. Each block is first DCT (discrete cosine transform) transformed, and then the DCT coefficients have to be quantized into integer. A zigzag scan or other alternate scanning method is utilized to arrange the two-dimensional quantized coefficients into one-dimensional. Finally, a variable-length coding (VLC) is employed to do entropy coding.
- VLC variable-length coding
- the quantization step is where the compression happened.
- the quantization of a DCT coefficient F[v] [u] can be represented as:
- v, u are the indexes for the two dimensional matrix, ranged from 0 to 7.
- the Q is the quantizer scale for the blocks within a macroblock, which can be varied for each macroblock.
- the W[v][u] is a quantization matrix defined for the whole pictures, which is used to assign a weighting factor for different DCT coefficients.
- FIG. 1 is a default quantization matrix for intra blocks defined in MPEG-2. Referring to FIG. 1, the W[v][u] is getting larger when the v and u indexes increase. This causes that the quantized coefficients with higher frequency more easily becomes zero. This is based on the study of human visual system that human is more sensitive to lower frequency signal and less sensitive to higher frequency.
- the inverse-quantization step is used to recover the DCT coefficients.
- the inverse-quantization of a quantized DCT coefficient QF[v][u] is defined as:
- F′[v][u] is the recovered DCT coefficient.
- quantization error which is defined as:
- the quantizer scale Q is used to scale the value of quantization matrix, ranged from 1 to 31.
- the adaptive quantization adapted in MPEG-1 and MPEG-2 can freely change the Q from one macroblock to the next macroblock, but can only slightly adjust the Q between adjacent macroblocks in MPEG-4.
- the MPEG-2 has a non-linear quantizer scale mapping the quantizer scale code (from 1 to 31) to a real quantizer scale (from 0.5 to 56).
- Adjusting the quantizer scale Q can control the bit consumption and coded quality of a macroblock. Given a larger Q will cause the quantized DCT coefficients getting smaller, and more quantized DCT coefficients became zero. Thus, the coded bit stream after VLC becomes shorter. However, the consequence is that the quantization error is getting larger, and the decoded image quality getting worse. If we want to have a better image quality, setting a smaller Q can reduce the quantization error, but the coded bit stream will getting longer.
- bit-rate of MPEG bit stream is constrained.
- DVD standard defines that the bit-rate of a MPEG-2 video stream can't higher than 9.8 Mb/sec, so the encoder have to control the bit-rate consumption to satisfy the constrain.
- the picture contents vary from picture to picture in a video sequence.
- the coding complexity in a portion of video sequence may be different to other portions.
- the video buffer verifier (hereinafter called VBV) employed in MPEG-1 can offer somewhat flexibility in bit-rate consumption for different pictures, but overall bit-rate in a long term should be a constant bit rate (hereinafter called CBR).
- CBR constant bit rate
- the MPEG-2 introduces the variable bit rate (VBR) operation mode to provide more flexibility on the variation of bit-rate consumption for each picture.
- VBV buffer is used to emulate the input buffer of the MPEG decoder, and the bit stream produced by a MPEG encoder can not violate the constraint on the VBV buffer, otherwise the bit stream wouldn't be properly decoded.
- the CBR operation of the VBV is shown by example in FIG. 2.
- the figure depicts the fullness of the decoder buffer over time.
- the sloped line segments show the compressed data entering the buffer at a constant bit-rate.
- the vertical line segments show the instantaneous removal from the buffer of the data associated with the decoded picture.
- the data for a picture must be already available in the buffer when the decoder has to decode the picture, and the decoder buffer does not overfill.
- the VBV buffer may be underflow, therefore the upper bound of the allocatable bits is UB.
- the VBV buffer may be overflow, and thus the lower bound of the allocatable bits is LB.
- the VBR operation of the VBV is shown by example in FIG. 3. The difference between VBR and CBR is that the compressed bit stream enters the buffer at a specified maximum bit-rate until the buffer is full, when no more bits are input. This translates to a bit-rate entering the buffer that may be effectively variable, up to the maximum specified rate. As shown in the FIG. 3, there is only a constraint on the allocatable maximal bit rate UB, but no minimal bit rate.
- the encoder has to allocate a bit budget for each picture, and then try to control the actually consumed bits close to the allocated bit budget.
- a virtual buffer mechanism is employed to control the bit consumption. If the buffer occupancy lower than zero, it maps to the quantizer scale 1 , and if the buffer occupancy higher than a threshold R, it maps to the quantizer scale 31 . The R is called the reaction factor.
- the virtual buffer occupancy D increases, otherwise the D decreases. If the buffer occupancy D increases over d0+R/31, the quantizer scale Q will become q0+1. This means that the excess of bit consumption is too large, so increase the quantizer scale to try to reduce the rate of bit consumption. If the buffer occupancy D decreases lower than d0 ⁇ R/31, the quantizer scale Q will became q0 ⁇ 1. This means that the bit consumption is lower than expected, so decrease the quantizer scale to try to increase the rate of bit consumption.
- the problem of using the virtual buffer mechanism to control the bit consumption is how to assign the initial buffer occupancy d0.
- This initial occupancy can be treat as an estimation of the coding complexity of the current picture. With a fixed bit budget, if the coding complexity of the current picture is relatively high, a higher initial occupancy should be selected so that the quantizer scale for each macroblock is higher. But we don't know the coding complexity of the current picture until the picture is actually coded.
- One solution is to inherit the buffer occupancy of the previous coded picture. This simple technique can handle most cases with no serious problem, but would cause VBV underflow when there is a rapidly variation of image content in a video sequence, such as senses change.
- the quantizer scale for a macroblock can be selected from 1 to 31; this dynamic range can handle the most cases of video coding. However, if a video sequence is very complex and the target bit rate is low, the VBV underflow seems to be unavoidable. Even if the quantizer scale for each macroblock is set to 31, there are still too many bits consumed for each macroblock.
- the U.S. Pat. No. 5,801,779 introduces a “Panic Mode” to overcome this situation. The basic idea is that encoder monitors the VBV buffer occupancy after coding each macroblock. If too many bits are consumed and there would be VBV underflow after coding the whole picture, the encoder enters the Panic Mode.
- the encoder would prefer to encode a macroblock with inter-mode, and no residuals are coded. If a macroblock is coded with intra-mode (The macroblocks of I-picture must be intra-mode), only a few numbers of coefficients are coded, making other coefficients zero to reduce the bits. The picture quality would degrade significantly, but the VBV underflow could be avoid.
- an object of the invention is to provide a quantization matrix adjusting method to avoid underflow of VBV and increase the picture quality by scaling up the quantization matrix.
- the quantization matrix adjusting method comprises the steps of: updating a used bits BB_X and an average quantizer scale Avg_Q; allocating the bit budget BB_C for the current picture; calculating an estimated quantizer scale EstQ_C according to the used bits BB_X, average quantizer scale Avg_Q and the bit budget BB_C; deciding the operation mode, when the estimated quantizer scale EstQ_C is larger than a threshold Th_Q, assigning the operation mode as an alert mode, otherwise remaining at normal mode; enlarging the default quantization matrix as a new quantization matrix when the operation mode is the alert mode; and coding the current picture, the encoder coding the current picture using the new quantization matrix and default quantization matrix in the alert mode and normal mode, respectively.
- FIG. 1 is a default quantization matrix for intra blocks defined in MPEG-2.
- FIG. 2 is the constant bit rate operation of the video buffer verifier.
- FIG. 3 is the variable bit rate operation of the video buffer verifier.
- FIG. 4 is a flow chart of quantization matrix adjusting method according to the first embodiment of present invention.
- FIG. 5 is a flow chart of quantization matrix adjusting method according to the second embodiment of present invention.
- the VBV underflow would typically occur when the current picture is encoded by the default quantization matrix and the maximal quantizer scale. Because the quantizer scale is ranged from 1 to 31, the encoder just can increase the weighting of quantization matrix to avoid the VBV underflow. So if encoder still wants to reduce the QF[v][u] while the Q is given the maximal value, increasing the W[v][u] is the only solution. However, changing quantization matrix has some constraint.
- the quantization matrixes will set to the value defined in the sequence header.
- the customized quantization matrix can only be set in the sequence header, and the sequence header can only inserted before a GOP header. This means that if we want to change the quantization matrix, we must start a new GOP, insert a sequence header before the GOP, and place the new quantization matrix in the sequence header.
- the quantization matrix can not only placed in the sequence header, but also can be changed for each picture by insert a quantization matrix extension. So if we want to change the quantization matrix, we can only do it at the beginning of encoding a picture.
- FIG. 4 is a flow chart of quantization matrix adjusting method according to the first embodiment of present invention.
- the encoder with the adjusting method of this embodiment encodes a current picture without a pre-scan to analysis the complexity of the current picture. Because the encoder encodes the current picture without a pre-scan to analysis the complexity of the current picture, like the well-known Test Model 5 (TM5) rate control algorithm, the encoder can only use the information of the previous coded pictures to estimate the coding complexity of the current picture.
- the method of the first embodiment of the present invention includes the following steps.
- Step S 402 update the used bits BB_X of the previously coded picture and the average quantizer scale Avg_Q.
- encoder can calculate the average quantizer scale Avg_Q, the final virtual buffer occupancy d_X of this picture and the used bits BB_X of the coded picture.
- Step S 404 allocate a bit budget BB_C for the current picture of type X.
- the initial virtual buffer occupancy d0_X of the next coding picture with type X is set to the final virtual buffer occupancy d_X, and the initial quantizer scale Q is set to the value corresponding to the average quantizer scale Avg_Q. Then, the frame-level rate control will allocate the bit budget BB_C for the current picture.
- Step S 406 calculate the estimated quantizer scale EstQ_C for the current picture.
- the encoder using the adjusting method of the first embodiment uses the relation of the used bits BB_X of the previously coded picture, the bit budget BB_C of the current picture and the average quantizer scale Avg_Q to calculate the estimated quantizer scale EstQ_C of the current picture.
- Equation (4) is one method to calculate the estimated quantizer scale EstQ_C.
- Step S 408 decide the operation mode. If the estimated quantizer scale EstQ_C is larger than a threshold ThQ, such as 28, 29, 30 or 31, the encoder will enter an alert mode to encode the current picture should. In the alert mode, it means that the picture content is pretty complex to be coded with the given bit rate. If the used bits BB_X of the previously coded picture and the bit budget BB_C assigned for the current picture are about the same, we should also believe that the current picture can't be easily coded with the given bit budget. Therefore, when the estimated quantizer scale EstQ_C is larger than the threshold ThQ, the encoder will jump to Step S 410 to enter the alert mode, otherwise jump to Step S 416 to remain the normal mode.
- a threshold ThQ such as 28, 29, 30 or 31
- Step 410 enlarge the quantization matrix as a new quantization matrix.
- the encoder wants the final average quantizer scale of the current picture reduced to a safer value SQ_X, which should smaller than or equal to threshold ThQ. Therefore, the encoder enlarges the quantization matrix by an adjustment factor S, and the initial virtual buffer occupancy will be set with the value corresponding to d0_X/S.
- the adjustment factor S should be a value larger than 1.
- the adjustment factor S is calculated as:
- the new quantization matrix is enlarged as:
- the W0[v][u] is the default quantization matrix or a user defined quantization matrix if desired. Therefore, the smaller the safer value SQ_X is, the larger the adjustment factor S is.
- the maximum value for the W′[v][u] is 255, so the new value has to cramp to smaller than or equal to 255.
- Step 412 encode the current picture using the new quantization matrix.
- Step 414 check the encoding state. If the current picture is encoded completely, jump back to Step S 402 , otherwise jump to Step S 412 .
- Step 416 encode the current picture using the default quantization matrix.
- Step 418 check the encoding state. If the current picture is encoded completely, jump back to Step S 402 , otherwise jump to Step S 416 .
- the quantization matrix will be enlarged by the adjustment factor S. Because some macroblocks are quantized using the new quantization matrix, the quantizer scale Q of these macroblocks should also be enlarged by adjustment factor S for calculating the Avg_Q to reflect the actual coding complexity. Remind that the quantizer scale Q used to quantize the macroblocks can not be larger than 31, but the enlarged Q used to calculate the Avg_Q would be larger than 31.
- the quantization matrix for these five adjacent pictures are all W2, and only the first picture has to encode the W2 into bit stream.
- the following four pictures can use the W2 without any effort until a sequence header is encountered.
- the sequence header will reset the quantization matrixes and the following first picture has to encode the quantization matrix for itself.
- FIG. 5 is a flow chart of quantization matrix adjusting method according to the second embodiment of present invention.
- the encoder with the adjusting method of the second embodiment encodes a current picture with a pre-scan to analysis the complexity of the current picture. If the encoder can do a pre-analysis on the current picture to collect the estimation of the complexity, a more correct decision can be made on whether to change the quantization and how to do it.
- the coding complexity of the current picture can be estimated while doing the motion estimation.
- the variance of each intra-coded macroblock can be obtained.
- a mode decision has to be made depends on the variances of the intra-coded and inter-coded mode.
- the encoder can only decide weather changes the quantization matrix based on the previously coded pictures. However, if there is a rapid change on the coding complexity of adjacent pictures, this decision may be failed. If the average quantizer scale Avg_Q of the previously coded picture is slightly smaller than the threshold ThQ, the current picture would not enter the alert mode. However, if the coding complexity of the current picture is much higher than the previously coded pictures, the final bit rate after coding the current picture would exceed the budget significantly, causing the VBV underflow occurred. In the contract, if the coding complexity of the previously coded picture is pretty high but the current picture is very low, an over-enlarged quantization matrix would cause the significantly degrade on the quality.
- the second embodiment of the present invention is to first estimate the activity Act_C of the current picture and then to estimate the estimated quantizer scale EstQ_C according to the activity Act_C.
- the encoder calculates the estimated quantizer scale EstQ_C of the current picture by the following equation:
- EstQ — C ( BB — X/BB — C )* Avg — Q* ( Act — C/Act — X ) (7)
- equation (7) is used to calculate the estimated quantizer scale EstQ_C by the parameters, including not only the bit budget BB_C, used bits BB_X, but also the activity Act_X and Act_C. That is, when the bit budget BB_C and average quantizer scale AvgQ_C of the current picture are same with that of previously picture, if the axtivity Act_C of the current picture is larger than the activity Act_X of previously picture, the estimated quantizer scale EstQ_C will be higher relatively to make the coded bit rate of the current picture not exceeding the bit budget much. In the contract, if the axtivity Act_C of the current picture is smaller than the activity Act_X of previously picture, the estimated quantizer scale EstQ_C will be lower relatively to make the coded bit rate of the current picture not under the bit budget much.
- the second embodiment is same with the first embodiment to decide whether the encoder enters the alert mode according to the estimated quantizer scale EstQ_C. That is, when the estimated quantizer scale EstQ_C is larger than the Threshold ThQ, the encoder enters the alert mode, otherwise remains the normal mode.
- the encoder When the encoder enters the alert mode, the encoder will enlarge the default quantization matrix as a new quantization matrix.
- the encoder wants the final average quantizer scale of the current picture reduced to a safer value SQ_X, which should smaller than or equal to threshold ThQ. Therefore, the encoder enlarges the quantization matrix by an adjustment factor S, and the initial virtual buffer occupancy will be set with the value corresponding to d0_X/S.
- the adjustment factor S should be a value larger than 1.0 and can be calculated by equation (5).
- the new quantization matrix can be calculated by equation (6).
- the maximum value for the W′[v][u] is 255, so the new value has to cramp to smaller than or equal to 255.
Abstract
Description
- 1. Field of the Invention
- The invention relates to a quantization matrix adjusting method, and more particularly to a quantization matrix adjusting method for avoiding underflow of video buffer verifier (VBV) by scaling up the quantization matrix.
- 2. Description of the Related Art
- Compressing methods are typically used to compress the pictures in order to reduce the data amounts, for example coding pictures by MPEG. The basic unit of coding within a picture is the macroblock. If the sampling is in 4:2:0 format, there are six blocks within a macroblock: four Y blocks, one Cb block and one Cr block. Each block is first DCT (discrete cosine transform) transformed, and then the DCT coefficients have to be quantized into integer. A zigzag scan or other alternate scanning method is utilized to arrange the two-dimensional quantized coefficients into one-dimensional. Finally, a variable-length coding (VLC) is employed to do entropy coding.
- The quantization step is where the compression happened. Generally, the quantization of a DCT coefficient F[v] [u] can be represented as:
- QF[v][u]=16*F[v][u]/(Q*W[v][u]) (1)
- Where the v, u are the indexes for the two dimensional matrix, ranged from 0 to 7. The Q is the quantizer scale for the blocks within a macroblock, which can be varied for each macroblock. The W[v][u] is a quantization matrix defined for the whole pictures, which is used to assign a weighting factor for different DCT coefficients. FIG. 1 is a default quantization matrix for intra blocks defined in MPEG-2. Referring to FIG. 1, the W[v][u] is getting larger when the v and u indexes increase. This causes that the quantized coefficients with higher frequency more easily becomes zero. This is based on the study of human visual system that human is more sensitive to lower frequency signal and less sensitive to higher frequency.
- The inverse-quantization step is used to recover the DCT coefficients. Generally, the inverse-quantization of a quantized DCT coefficient QF[v][u] is defined as:
- F′[v][u]=QF[v][u]*Q*W[v][u]/16 (2)
- where F′[v][u] is the recovered DCT coefficient. The difference between the original DCT coefficient F[v][u] and the inverse-quantized DCT coefficient F′[v][u] is called quantization error, which is defined as:
- E[v][u]=F[v][u]−F′[v][u] (3)
- The quantizer scale Q is used to scale the value of quantization matrix, ranged from 1 to 31. The adaptive quantization adapted in MPEG-1 and MPEG-2 can freely change the Q from one macroblock to the next macroblock, but can only slightly adjust the Q between adjacent macroblocks in MPEG-4. The MPEG-2 has a non-linear quantizer scale mapping the quantizer scale code (from 1 to 31) to a real quantizer scale (from 0.5 to 56).
- Adjusting the quantizer scale Q can control the bit consumption and coded quality of a macroblock. Given a larger Q will cause the quantized DCT coefficients getting smaller, and more quantized DCT coefficients became zero. Thus, the coded bit stream after VLC becomes shorter. However, the consequence is that the quantization error is getting larger, and the decoded image quality getting worse. If we want to have a better image quality, setting a smaller Q can reduce the quantization error, but the coded bit stream will getting longer.
- In typical applications, the bit-rate of MPEG bit stream is constrained. For example, the DVD standard defines that the bit-rate of a MPEG-2 video stream can't higher than 9.8 Mb/sec, so the encoder have to control the bit-rate consumption to satisfy the constrain.
- The picture contents vary from picture to picture in a video sequence. The coding complexity in a portion of video sequence may be different to other portions. The video buffer verifier (hereinafter called VBV) employed in MPEG-1 can offer somewhat flexibility in bit-rate consumption for different pictures, but overall bit-rate in a long term should be a constant bit rate (hereinafter called CBR). The MPEG-2 introduces the variable bit rate (VBR) operation mode to provide more flexibility on the variation of bit-rate consumption for each picture. The VBV buffer is used to emulate the input buffer of the MPEG decoder, and the bit stream produced by a MPEG encoder can not violate the constraint on the VBV buffer, otherwise the bit stream wouldn't be properly decoded. The CBR operation of the VBV is shown by example in FIG. 2. The figure depicts the fullness of the decoder buffer over time. The sloped line segments show the compressed data entering the buffer at a constant bit-rate. The vertical line segments show the instantaneous removal from the buffer of the data associated with the decoded picture.
- For the bit stream to satisfy the MPEG rate control requirement, the data for a picture must be already available in the buffer when the decoder has to decode the picture, and the decoder buffer does not overfill. By referencing to FIG. 2, if a picture consumed too many bits, the VBV buffer may be underflow, therefore the upper bound of the allocatable bits is UB. Similarly, if a picture used too few bits, the VBV buffer may be overflow, and thus the lower bound of the allocatable bits is LB. The VBR operation of the VBV is shown by example in FIG. 3. The difference between VBR and CBR is that the compressed bit stream enters the buffer at a specified maximum bit-rate until the buffer is full, when no more bits are input. This translates to a bit-rate entering the buffer that may be effectively variable, up to the maximum specified rate. As shown in the FIG. 3, there is only a constraint on the allocatable maximal bit rate UB, but no minimal bit rate.
- To satisfy the constraint of VBV, the encoder has to allocate a bit budget for each picture, and then try to control the actually consumed bits close to the allocated bit budget. Typically, a virtual buffer mechanism is employed to control the bit consumption. If the buffer occupancy lower than zero, it maps to the
quantizer scale 1, and if the buffer occupancy higher than a threshold R, it maps to the quantizer scale 31. The R is called the reaction factor. Before coding a picture, the virtual buffer has an initial occupancy D=d0=R*q0/31, and this occupancy corresponds to a quantizer scale Q=q0. Use the Q to quantize the first macroblock and encode. If the consumed bits are larger than the average bit budget for a macroblock, the virtual buffer occupancy D increases, otherwise the D decreases. If the buffer occupancy D increases over d0+R/31, the quantizer scale Q will become q0+1. This means that the excess of bit consumption is too large, so increase the quantizer scale to try to reduce the rate of bit consumption. If the buffer occupancy D decreases lower than d0−R/31, the quantizer scale Q will became q0−1. This means that the bit consumption is lower than expected, so decrease the quantizer scale to try to increase the rate of bit consumption. - The problem of using the virtual buffer mechanism to control the bit consumption is how to assign the initial buffer occupancy d0. This initial occupancy can be treat as an estimation of the coding complexity of the current picture. With a fixed bit budget, if the coding complexity of the current picture is relatively high, a higher initial occupancy should be selected so that the quantizer scale for each macroblock is higher. But we don't know the coding complexity of the current picture until the picture is actually coded. One solution is to inherit the buffer occupancy of the previous coded picture. This simple technique can handle most cases with no serious problem, but would cause VBV underflow when there is a rapidly variation of image content in a video sequence, such as senses change.
- The quantizer scale for a macroblock can be selected from 1 to 31; this dynamic range can handle the most cases of video coding. However, if a video sequence is very complex and the target bit rate is low, the VBV underflow seems to be unavoidable. Even if the quantizer scale for each macroblock is set to 31, there are still too many bits consumed for each macroblock. The U.S. Pat. No. 5,801,779 introduces a “Panic Mode” to overcome this situation. The basic idea is that encoder monitors the VBV buffer occupancy after coding each macroblock. If too many bits are consumed and there would be VBV underflow after coding the whole picture, the encoder enters the Panic Mode. In this mode, the encoder would prefer to encode a macroblock with inter-mode, and no residuals are coded. If a macroblock is coded with intra-mode (The macroblocks of I-picture must be intra-mode), only a few numbers of coefficients are coded, making other coefficients zero to reduce the bits. The picture quality would degrade significantly, but the VBV underflow could be avoid.
- In view of the above-mentioned problems, an object of the invention is to provide a quantization matrix adjusting method to avoid underflow of VBV and increase the picture quality by scaling up the quantization matrix.
- To achieve the above-mentioned object, the quantization matrix adjusting method comprises the steps of: updating a used bits BB_X and an average quantizer scale Avg_Q; allocating the bit budget BB_C for the current picture; calculating an estimated quantizer scale EstQ_C according to the used bits BB_X, average quantizer scale Avg_Q and the bit budget BB_C; deciding the operation mode, when the estimated quantizer scale EstQ_C is larger than a threshold Th_Q, assigning the operation mode as an alert mode, otherwise remaining at normal mode; enlarging the default quantization matrix as a new quantization matrix when the operation mode is the alert mode; and coding the current picture, the encoder coding the current picture using the new quantization matrix and default quantization matrix in the alert mode and normal mode, respectively.
- FIG. 1 is a default quantization matrix for intra blocks defined in MPEG-2.
- FIG. 2 is the constant bit rate operation of the video buffer verifier.
- FIG. 3 is the variable bit rate operation of the video buffer verifier.
- FIG. 4 is a flow chart of quantization matrix adjusting method according to the first embodiment of present invention.
- FIG. 5 is a flow chart of quantization matrix adjusting method according to the second embodiment of present invention.
- The quantization matrix adjusting method for avoiding underflow of video buffer verifier of the present invention will be described with reference to the accompanying drawings.
- The VBV underflow would typically occur when the current picture is encoded by the default quantization matrix and the maximal quantizer scale. Because the quantizer scale is ranged from 1 to 31, the encoder just can increase the weighting of quantization matrix to avoid the VBV underflow. So if encoder still wants to reduce the QF[v][u] while the Q is given the maximal value, increasing the W[v][u] is the only solution. However, changing quantization matrix has some constraint.
- In MPEG-1 and MPEG-2, there must be a sequence header placed at the beginning of coded bit stream, and the sequence header can be added in the subsequence of bit stream. Once a sequence header is encountered, the quantization matrixes will set to the value defined in the sequence header. In MPEG-1, the customized quantization matrix can only be set in the sequence header, and the sequence header can only inserted before a GOP header. This means that if we want to change the quantization matrix, we must start a new GOP, insert a sequence header before the GOP, and place the new quantization matrix in the sequence header. In the MPEG-2, the quantization matrix can not only placed in the sequence header, but also can be changed for each picture by insert a quantization matrix extension. So if we want to change the quantization matrix, we can only do it at the beginning of encoding a picture.
- FIG. 4 is a flow chart of quantization matrix adjusting method according to the first embodiment of present invention. The encoder with the adjusting method of this embodiment encodes a current picture without a pre-scan to analysis the complexity of the current picture. Because the encoder encodes the current picture without a pre-scan to analysis the complexity of the current picture, like the well-known Test Model 5 (TM5) rate control algorithm, the encoder can only use the information of the previous coded pictures to estimate the coding complexity of the current picture. The method of the first embodiment of the present invention includes the following steps.
- Step S402: update the used bits BB_X of the previously coded picture and the average quantizer scale Avg_Q. After coding a picture with type of X (X is in I-, P- or B-picture), encoder can calculate the average quantizer scale Avg_Q, the final virtual buffer occupancy d_X of this picture and the used bits BB_X of the coded picture.
- Step S404: allocate a bit budget BB_C for the current picture of type X. In the general cases, the initial virtual buffer occupancy d0_X of the next coding picture with type X is set to the final virtual buffer occupancy d_X, and the initial quantizer scale Q is set to the value corresponding to the average quantizer scale Avg_Q. Then, the frame-level rate control will allocate the bit budget BB_C for the current picture.
- Step S406: calculate the estimated quantizer scale EstQ_C for the current picture. By assuming the activity would be almost the same between adjacent pictures of the same type, the encoder using the adjusting method of the first embodiment uses the relation of the used bits BB_X of the previously coded picture, the bit budget BB_C of the current picture and the average quantizer scale Avg_Q to calculate the estimated quantizer scale EstQ_C of the current picture. Equation (4) is one method to calculate the estimated quantizer scale EstQ_C.
- EstQ — C(BB — X/BB — C)*Avg — Q (4)
- Step S408: decide the operation mode. If the estimated quantizer scale EstQ_C is larger than a threshold ThQ, such as 28, 29, 30 or 31, the encoder will enter an alert mode to encode the current picture should. In the alert mode, it means that the picture content is pretty complex to be coded with the given bit rate. If the used bits BB_X of the previously coded picture and the bit budget BB_C assigned for the current picture are about the same, we should also believe that the current picture can't be easily coded with the given bit budget. Therefore, when the estimated quantizer scale EstQ_C is larger than the threshold ThQ, the encoder will jump to Step S410 to enter the alert mode, otherwise jump to Step S416 to remain the normal mode.
- Step410: enlarge the quantization matrix as a new quantization matrix. In the alert mode, the encoder wants the final average quantizer scale of the current picture reduced to a safer value SQ_X, which should smaller than or equal to threshold ThQ. Therefore, the encoder enlarges the quantization matrix by an adjustment factor S, and the initial virtual buffer occupancy will be set with the value corresponding to d0_X/S. The adjustment factor S should be a value larger than 1. The adjustment factor S is calculated as:
- S=EstQ — C/SQ — X (5)
- The new quantization matrix is enlarged as:
- W′[v] [u]=W0[v] [u]*S (6)
- where v=0 . . . 7, u=0 . . . 7. The W0[v][u] is the default quantization matrix or a user defined quantization matrix if desired. Therefore, the smaller the safer value SQ_X is, the larger the adjustment factor S is. The maximum value for the W′[v][u] is 255, so the new value has to cramp to smaller than or equal to 255.
- Usually, there are two default quantization matrixes in the encoder, one for intra-block and one for inter-block. Update a new quantization matrix requires 512 bits to encode the quantization matrix itself and some extra bits, so the number of updating quantization matrix would be kept as small as possible. Only intra quantization matrix is required for I-picture, and maybe only updating the inter quantization matrix is enough for the B-picture. In the case of P-picture, it would depend on how many intra-coded macroblocks existed in the picture. If the number of intra-coded macroblock is larger than a threshold, updating two quantization matrixes would be a better choice. Otherwise, only the inter quantization matrix has to be updated.
- Step412: encode the current picture using the new quantization matrix.
- Step414: check the encoding state. If the current picture is encoded completely, jump back to Step S402, otherwise jump to Step S412.
- Step416: encode the current picture using the default quantization matrix.
- Step418: check the encoding state. If the current picture is encoded completely, jump back to Step S402, otherwise jump to Step S416.
- In the above steps, if the operation mode is the alert mode, the quantization matrix will be enlarged by the adjustment factor S. Because some macroblocks are quantized using the new quantization matrix, the quantizer scale Q of these macroblocks should also be enlarged by adjustment factor S for calculating the Avg_Q to reflect the actual coding complexity. Remind that the quantizer scale Q used to quantize the macroblocks can not be larger than 31, but the enlarged Q used to calculate the Avg_Q would be larger than 31.
- If the values of the adjustment factor S for some adjacent pictures are slightly different, changing the quantization matrix for every picture to reflect the small difference is not a good idea. It is a better choice to change the quantization matrix when the difference is larger than a threshold. Another solution is to “quantize” the adjustment factor S so that there are only some pre-defined quantization matrixes corresponding to some possible quantized adjustment factor S′. Quantize the adjustment factor S to the most closed adjustment factor S′ and then used the corresponding quantization matrix. For example, if there are four possible adjustment factor S′: 1.1, 1.2, 1.3 and 1.4, and the default quantization matrix is enlarged by these four value to get four new quantization matrix W1, W2, W3 and W4, respectively. If the adjustment factor S for five adjacent pictures are 1.21, 1.18, 1.16, 1.22 and 1.2, the quantization matrix for these five adjacent pictures are all W2, and only the first picture has to encode the W2 into bit stream. The following four pictures can use the W2 without any effort until a sequence header is encountered. The sequence header will reset the quantization matrixes and the following first picture has to encode the quantization matrix for itself.
- FIG. 5 is a flow chart of quantization matrix adjusting method according to the second embodiment of present invention. The encoder with the adjusting method of the second embodiment encodes a current picture with a pre-scan to analysis the complexity of the current picture. If the encoder can do a pre-analysis on the current picture to collect the estimation of the complexity, a more correct decision can be made on whether to change the quantization and how to do it. Typically, the coding complexity of the current picture can be estimated while doing the motion estimation. For the I-picture, the variance of each intra-coded macroblock can be obtained. For the P- and B-pictures, a mode decision has to be made depends on the variances of the intra-coded and inter-coded mode. We call the sum of variances as the activities of the current pictures. Before coding a picture, we know the activity Act_C of the current picture. After coding a picture, the average quantizer scale Avg_Q and the actually used bits BB_X can be used to calculate the complexity Com of the current picture.
- In the encoder without pre-analysis function, the encoder can only decide weather changes the quantization matrix based on the previously coded pictures. However, if there is a rapid change on the coding complexity of adjacent pictures, this decision may be failed. If the average quantizer scale Avg_Q of the previously coded picture is slightly smaller than the threshold ThQ, the current picture would not enter the alert mode. However, if the coding complexity of the current picture is much higher than the previously coded pictures, the final bit rate after coding the current picture would exceed the budget significantly, causing the VBV underflow occurred. In the contract, if the coding complexity of the previously coded picture is pretty high but the current picture is very low, an over-enlarged quantization matrix would cause the significantly degrade on the quality.
- The difference between the first and second embodiments is that the second embodiment of the present invention is to first estimate the activity Act_C of the current picture and then to estimate the estimated quantizer scale EstQ_C according to the activity Act_C. The encoder calculates the estimated quantizer scale EstQ_C of the current picture by the following equation:
- EstQ — C=(BB — X/BB — C)*Avg — Q*(Act — C/Act — X) (7)
- The difference between equation (7) and equation (4) is that equation (7) is used to calculate the estimated quantizer scale EstQ_C by the parameters, including not only the bit budget BB_C, used bits BB_X, but also the activity Act_X and Act_C. That is, when the bit budget BB_C and average quantizer scale AvgQ_C of the current picture are same with that of previously picture, if the axtivity Act_C of the current picture is larger than the activity Act_X of previously picture, the estimated quantizer scale EstQ_C will be higher relatively to make the coded bit rate of the current picture not exceeding the bit budget much. In the contract, if the axtivity Act_C of the current picture is smaller than the activity Act_X of previously picture, the estimated quantizer scale EstQ_C will be lower relatively to make the coded bit rate of the current picture not under the bit budget much.
- After calculating the estimated quantizer scale EstQ_C, the second embodiment is same with the first embodiment to decide whether the encoder enters the alert mode according to the estimated quantizer scale EstQ_C. That is, when the estimated quantizer scale EstQ_C is larger than the Threshold ThQ, the encoder enters the alert mode, otherwise remains the normal mode.
- When the encoder enters the alert mode, the encoder will enlarge the default quantization matrix as a new quantization matrix. In the alert mode, the encoder wants the final average quantizer scale of the current picture reduced to a safer value SQ_X, which should smaller than or equal to threshold ThQ. Therefore, the encoder enlarges the quantization matrix by an adjustment factor S, and the initial virtual buffer occupancy will be set with the value corresponding to d0_X/S. The adjustment factor S should be a value larger than 1.0 and can be calculated by equation (5). The new quantization matrix can be calculated by equation (6). The maximum value for the W′[v][u] is 255, so the new value has to cramp to smaller than or equal to 255.
- While certain exemplary embodiments have been described and shown in the accompanying drawings, it is to be understood that such embodiments are merely illustrative of and not restrictive on the broad invention, and that this invention should not be limited to the specific construction and arrangement shown and described, since various other modifications may occur to those ordinarily skilled in the art.
Claims (14)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US10/442,280 US7388995B2 (en) | 2003-05-21 | 2003-05-21 | Quantization matrix adjusting method for avoiding underflow of data |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US10/442,280 US7388995B2 (en) | 2003-05-21 | 2003-05-21 | Quantization matrix adjusting method for avoiding underflow of data |
Publications (2)
Publication Number | Publication Date |
---|---|
US20040234150A1 true US20040234150A1 (en) | 2004-11-25 |
US7388995B2 US7388995B2 (en) | 2008-06-17 |
Family
ID=33450157
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US10/442,280 Expired - Fee Related US7388995B2 (en) | 2003-05-21 | 2003-05-21 | Quantization matrix adjusting method for avoiding underflow of data |
Country Status (1)
Country | Link |
---|---|
US (1) | US7388995B2 (en) |
Cited By (24)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050152449A1 (en) * | 2004-01-12 | 2005-07-14 | Nemiroff Robert S. | Method and apparatus for processing a bitstream in a digital video transcoder |
US20100128781A1 (en) * | 2007-07-18 | 2010-05-27 | Jang Euee-Seon | Method and apparatus for coding and decoding using bit-precision |
US20100316120A1 (en) * | 2009-06-15 | 2010-12-16 | Kiyofumi Abe | Moving picture coding device and broadcast wave recording device |
US7912122B2 (en) | 2004-01-20 | 2011-03-22 | Panasonic Corporation | Picture coding method, picture decoding method, picture coding apparatus, picture decoding apparatus |
US20140369617A1 (en) * | 2013-06-13 | 2014-12-18 | Canon Kabushiki Kaisha | Image encoding apparatus, image encoding method, and program |
US9092353B1 (en) | 2013-01-29 | 2015-07-28 | Pmc-Sierra Us, Inc. | Apparatus and method based on LDPC codes for adjusting a correctable raw bit error rate limit in a memory system |
US9113164B1 (en) * | 2012-05-15 | 2015-08-18 | Google Inc. | Constant bit rate control using implicit quantization values |
US9128858B1 (en) * | 2013-01-29 | 2015-09-08 | Pmc-Sierra Us, Inc. | Apparatus and method for adjusting a correctable raw bit error rate limit in a memory system using strong log-likelihood (LLR) values |
US9397701B1 (en) | 2013-03-11 | 2016-07-19 | Microsemi Storage Solutions (Us), Inc. | System and method for lifetime specific LDPC decoding |
US9417804B2 (en) | 2014-07-07 | 2016-08-16 | Microsemi Storage Solutions (Us), Inc. | System and method for memory block pool wear leveling |
US9450610B1 (en) | 2013-03-15 | 2016-09-20 | Microsemi Storage Solutions (Us), Inc. | High quality log likelihood ratios determined using two-index look-up table |
US9454414B2 (en) | 2013-03-15 | 2016-09-27 | Microsemi Storage Solutions (Us), Inc. | System and method for accumulating soft information in LDPC decoding |
US9590656B2 (en) | 2013-03-15 | 2017-03-07 | Microsemi Storage Solutions (Us), Inc. | System and method for higher quality log likelihood ratios in LDPC decoding |
US9799405B1 (en) | 2015-07-29 | 2017-10-24 | Ip Gem Group, Llc | Nonvolatile memory system with read circuit for performing reads using threshold voltage shift read instruction |
US9813080B1 (en) | 2013-03-05 | 2017-11-07 | Microsemi Solutions (U.S.), Inc. | Layer specific LDPC decoder |
US9886214B2 (en) | 2015-12-11 | 2018-02-06 | Ip Gem Group, Llc | Nonvolatile memory system with erase suspend circuit and method for erase suspend management |
US9892794B2 (en) | 2016-01-04 | 2018-02-13 | Ip Gem Group, Llc | Method and apparatus with program suspend using test mode |
US9899092B2 (en) | 2016-01-27 | 2018-02-20 | Ip Gem Group, Llc | Nonvolatile memory system with program step manager and method for program step management |
US10157677B2 (en) | 2016-07-28 | 2018-12-18 | Ip Gem Group, Llc | Background reference positioning and local reference positioning using threshold voltage shift read |
US10171807B2 (en) * | 2015-01-29 | 2019-01-01 | Arris Enterprises Llc | Picture-level QP rate control for HEVC encoding |
US10230396B1 (en) | 2013-03-05 | 2019-03-12 | Microsemi Solutions (Us), Inc. | Method and apparatus for layer-specific LDPC decoding |
US10236915B2 (en) | 2016-07-29 | 2019-03-19 | Microsemi Solutions (U.S.), Inc. | Variable T BCH encoding |
US10291263B2 (en) | 2016-07-28 | 2019-05-14 | Ip Gem Group, Llc | Auto-learning log likelihood ratio |
US10332613B1 (en) | 2015-05-18 | 2019-06-25 | Microsemi Solutions (Us), Inc. | Nonvolatile memory system with retention monitor |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP4704972B2 (en) * | 2006-07-24 | 2011-06-22 | ルネサスエレクトロニクス株式会社 | Stream editing method and stream editing apparatus |
TW201034465A (en) * | 2009-03-11 | 2010-09-16 | Jmicron Technology Corp | Image compression method and related apparatus thereof |
Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5323187A (en) * | 1991-12-20 | 1994-06-21 | Samsung Electronics Co., Ltd. | Image compression system by setting fixed bit rates |
US5404174A (en) * | 1992-06-29 | 1995-04-04 | Victor Company Of Japan, Ltd. | Scene change detector for detecting a scene change of a moving picture |
US5801779A (en) * | 1995-12-26 | 1998-09-01 | C-Cube Microsystems, Inc. | Rate control with panic mode |
US5933532A (en) * | 1996-08-30 | 1999-08-03 | Sony Corporation | Video data compression apparatus and method of same |
US6026190A (en) * | 1994-10-31 | 2000-02-15 | Intel Corporation | Image signal encoding with variable low-pass filter |
US6111991A (en) * | 1998-01-16 | 2000-08-29 | Sharp Laboratories Of America | Method and apparatus for optimizing quantizer values in an image encoder |
US6222881B1 (en) * | 1994-10-18 | 2001-04-24 | Intel Corporation | Using numbers of non-zero quantized transform signals and signal differences to determine when to encode video signals using inter-frame or intra-frame encoding |
US6259739B1 (en) * | 1996-11-26 | 2001-07-10 | Matsushita Electric Industrial Co., Ltd. | Moving picture variable bit rate coding apparatus, moving picture variable bit rate coding method, and recording medium for moving picture variable bit rate coding program |
US6678322B1 (en) * | 1997-07-08 | 2004-01-13 | Sony Corporation | Video data coding device, video data coding method, video data transmitting device, and video data recording medium |
US6731685B1 (en) * | 2000-09-20 | 2004-05-04 | General Instrument Corporation | Method and apparatus for determining a bit rate need parameter in a statistical multiplexer |
US6763138B1 (en) * | 1999-03-11 | 2004-07-13 | Nec Corporation | Method and apparatus for coding moving picture at variable bit rate |
US6915018B2 (en) * | 1999-12-24 | 2005-07-05 | Nec Corporation | Compressed moving picture re-encoding apparatus and compressed moving picture re-encoding method |
US6963608B1 (en) * | 1998-10-02 | 2005-11-08 | General Instrument Corporation | Method and apparatus for providing rate control in a video encoder |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6345673B1 (en) * | 1998-11-20 | 2002-02-12 | Smith International, Inc. | High offset bits with super-abrasive cutters |
-
2003
- 2003-05-21 US US10/442,280 patent/US7388995B2/en not_active Expired - Fee Related
Patent Citations (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5323187A (en) * | 1991-12-20 | 1994-06-21 | Samsung Electronics Co., Ltd. | Image compression system by setting fixed bit rates |
US5404174A (en) * | 1992-06-29 | 1995-04-04 | Victor Company Of Japan, Ltd. | Scene change detector for detecting a scene change of a moving picture |
US6222881B1 (en) * | 1994-10-18 | 2001-04-24 | Intel Corporation | Using numbers of non-zero quantized transform signals and signal differences to determine when to encode video signals using inter-frame or intra-frame encoding |
US6026190A (en) * | 1994-10-31 | 2000-02-15 | Intel Corporation | Image signal encoding with variable low-pass filter |
US5801779A (en) * | 1995-12-26 | 1998-09-01 | C-Cube Microsystems, Inc. | Rate control with panic mode |
US5933532A (en) * | 1996-08-30 | 1999-08-03 | Sony Corporation | Video data compression apparatus and method of same |
US6259739B1 (en) * | 1996-11-26 | 2001-07-10 | Matsushita Electric Industrial Co., Ltd. | Moving picture variable bit rate coding apparatus, moving picture variable bit rate coding method, and recording medium for moving picture variable bit rate coding program |
US6678322B1 (en) * | 1997-07-08 | 2004-01-13 | Sony Corporation | Video data coding device, video data coding method, video data transmitting device, and video data recording medium |
US6111991A (en) * | 1998-01-16 | 2000-08-29 | Sharp Laboratories Of America | Method and apparatus for optimizing quantizer values in an image encoder |
US6963608B1 (en) * | 1998-10-02 | 2005-11-08 | General Instrument Corporation | Method and apparatus for providing rate control in a video encoder |
US6763138B1 (en) * | 1999-03-11 | 2004-07-13 | Nec Corporation | Method and apparatus for coding moving picture at variable bit rate |
US6915018B2 (en) * | 1999-12-24 | 2005-07-05 | Nec Corporation | Compressed moving picture re-encoding apparatus and compressed moving picture re-encoding method |
US6731685B1 (en) * | 2000-09-20 | 2004-05-04 | General Instrument Corporation | Method and apparatus for determining a bit rate need parameter in a statistical multiplexer |
Cited By (32)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050152449A1 (en) * | 2004-01-12 | 2005-07-14 | Nemiroff Robert S. | Method and apparatus for processing a bitstream in a digital video transcoder |
US7675970B2 (en) * | 2004-01-12 | 2010-03-09 | General Instrument Corporation | Method and apparatus for processing a bitstream in a digital video transcoder |
US7912122B2 (en) | 2004-01-20 | 2011-03-22 | Panasonic Corporation | Picture coding method, picture decoding method, picture coding apparatus, picture decoding apparatus |
US20110110423A1 (en) * | 2004-01-20 | 2011-05-12 | Shinya Kadono | Picture coding method, picture decoding method, picture coding apparatus, picture decoding apparatus, and program thereof |
US7995650B2 (en) | 2004-01-20 | 2011-08-09 | Panasonic Corporation | Picture coding method, picture decoding method, picture coding apparatus, picture decoding apparatus, and program thereof |
US8675734B2 (en) * | 2007-07-18 | 2014-03-18 | Humax Co., Ltd. | Method and apparatus for coding and decoding using bit-precision |
US20100128781A1 (en) * | 2007-07-18 | 2010-05-27 | Jang Euee-Seon | Method and apparatus for coding and decoding using bit-precision |
US20100316120A1 (en) * | 2009-06-15 | 2010-12-16 | Kiyofumi Abe | Moving picture coding device and broadcast wave recording device |
US9113164B1 (en) * | 2012-05-15 | 2015-08-18 | Google Inc. | Constant bit rate control using implicit quantization values |
US9092353B1 (en) | 2013-01-29 | 2015-07-28 | Pmc-Sierra Us, Inc. | Apparatus and method based on LDPC codes for adjusting a correctable raw bit error rate limit in a memory system |
US9128858B1 (en) * | 2013-01-29 | 2015-09-08 | Pmc-Sierra Us, Inc. | Apparatus and method for adjusting a correctable raw bit error rate limit in a memory system using strong log-likelihood (LLR) values |
US9448881B1 (en) | 2013-01-29 | 2016-09-20 | Microsemi Storage Solutions (Us), Inc. | Memory controller and integrated circuit device for correcting errors in data read from memory cells |
US9813080B1 (en) | 2013-03-05 | 2017-11-07 | Microsemi Solutions (U.S.), Inc. | Layer specific LDPC decoder |
US10230396B1 (en) | 2013-03-05 | 2019-03-12 | Microsemi Solutions (Us), Inc. | Method and apparatus for layer-specific LDPC decoding |
US9397701B1 (en) | 2013-03-11 | 2016-07-19 | Microsemi Storage Solutions (Us), Inc. | System and method for lifetime specific LDPC decoding |
US9450610B1 (en) | 2013-03-15 | 2016-09-20 | Microsemi Storage Solutions (Us), Inc. | High quality log likelihood ratios determined using two-index look-up table |
US9454414B2 (en) | 2013-03-15 | 2016-09-27 | Microsemi Storage Solutions (Us), Inc. | System and method for accumulating soft information in LDPC decoding |
US9590656B2 (en) | 2013-03-15 | 2017-03-07 | Microsemi Storage Solutions (Us), Inc. | System and method for higher quality log likelihood ratios in LDPC decoding |
US20140369617A1 (en) * | 2013-06-13 | 2014-12-18 | Canon Kabushiki Kaisha | Image encoding apparatus, image encoding method, and program |
US10123021B2 (en) * | 2013-06-13 | 2018-11-06 | Canon Kabushiki Kaisha | Image encoding apparatus for determining quantization parameter, image encoding method, and program |
US9417804B2 (en) | 2014-07-07 | 2016-08-16 | Microsemi Storage Solutions (Us), Inc. | System and method for memory block pool wear leveling |
US10171807B2 (en) * | 2015-01-29 | 2019-01-01 | Arris Enterprises Llc | Picture-level QP rate control for HEVC encoding |
US10332613B1 (en) | 2015-05-18 | 2019-06-25 | Microsemi Solutions (Us), Inc. | Nonvolatile memory system with retention monitor |
US9799405B1 (en) | 2015-07-29 | 2017-10-24 | Ip Gem Group, Llc | Nonvolatile memory system with read circuit for performing reads using threshold voltage shift read instruction |
US9886214B2 (en) | 2015-12-11 | 2018-02-06 | Ip Gem Group, Llc | Nonvolatile memory system with erase suspend circuit and method for erase suspend management |
US10152273B2 (en) | 2015-12-11 | 2018-12-11 | Ip Gem Group, Llc | Nonvolatile memory controller and method for erase suspend management that increments the number of program and erase cycles after erase suspend |
US9892794B2 (en) | 2016-01-04 | 2018-02-13 | Ip Gem Group, Llc | Method and apparatus with program suspend using test mode |
US9899092B2 (en) | 2016-01-27 | 2018-02-20 | Ip Gem Group, Llc | Nonvolatile memory system with program step manager and method for program step management |
US10157677B2 (en) | 2016-07-28 | 2018-12-18 | Ip Gem Group, Llc | Background reference positioning and local reference positioning using threshold voltage shift read |
US10283215B2 (en) | 2016-07-28 | 2019-05-07 | Ip Gem Group, Llc | Nonvolatile memory system with background reference positioning and local reference positioning |
US10291263B2 (en) | 2016-07-28 | 2019-05-14 | Ip Gem Group, Llc | Auto-learning log likelihood ratio |
US10236915B2 (en) | 2016-07-29 | 2019-03-19 | Microsemi Solutions (U.S.), Inc. | Variable T BCH encoding |
Also Published As
Publication number | Publication date |
---|---|
US7388995B2 (en) | 2008-06-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US7388995B2 (en) | Quantization matrix adjusting method for avoiding underflow of data | |
US6587508B1 (en) | Apparatus, method and computer program product for transcoding a coded moving picture sequence | |
US7653129B2 (en) | Method and apparatus for providing intra coding frame bit budget | |
US7426309B2 (en) | Method of controlling encoding rate, method of transmitting video data, encoding rate controller for video encoder, and video data transmission system using the encoding rate controller | |
US9369720B2 (en) | Image coding apparatus and image coding method | |
US6658157B1 (en) | Method and apparatus for converting image information | |
US7925108B2 (en) | Encoding device and dynamic image recording system having the encoding device | |
US7095784B2 (en) | Method and apparatus for moving picture compression rate control using bit allocation with initial quantization step size estimation at picture level | |
US20070177665A1 (en) | Picture layer rate control for video encoding | |
JP2001169281A (en) | Device and method for encoding moving image | |
WO2010041488A1 (en) | Dynamic image encoding device | |
WO2002032141A2 (en) | Video bitrate control method and device for digital video recording | |
JP4532980B2 (en) | Moving picture coding apparatus and method, computer program, and computer-readable storage medium | |
US20040234142A1 (en) | Apparatus for constant quality rate control in video compression and target bit allocator thereof | |
JPH09200758A (en) | Image encoder | |
US8081679B2 (en) | Image processing apparatus | |
JPH09322176A (en) | Encoding mode selecting method, moving image encoding device, encoding method, recording method and transmitting method | |
US6865224B1 (en) | Bit rate control for video data compression | |
JPH10336654A (en) | Image encoder | |
KR100950525B1 (en) | Transcoding apparatus and method, target bit allocation, complexity prediction apparatus and method of picture therein | |
US7391916B2 (en) | Quantization matrix adjusting method for quality improvement | |
US7133448B2 (en) | Method and apparatus for rate control in moving picture video compression | |
JPH09149412A (en) | Video data compression method | |
JPH07203430A (en) | Image coding device | |
JP4718736B2 (en) | Video encoding device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: SILICON INTEGRATED SYSTEMS CORP., TAIWAN Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:CHANG, YUNG-CHING;REEL/FRAME:014104/0194 Effective date: 20030506 |
|
STCF | Information on status: patent grant |
Free format text: PATENTED CASE |
|
FPAY | Fee payment |
Year of fee payment: 4 |
|
FEPP | Fee payment procedure |
Free format text: PAT HOLDER CLAIMS SMALL ENTITY STATUS, ENTITY STATUS SET TO SMALL (ORIGINAL EVENT CODE: LTOS); ENTITY STATUS OF PATENT OWNER: SMALL ENTITY |
|
FPAY | Fee payment |
Year of fee payment: 8 |
|
FEPP | Fee payment procedure |
Free format text: MAINTENANCE FEE REMINDER MAILED (ORIGINAL EVENT CODE: REM.); ENTITY STATUS OF PATENT OWNER: SMALL ENTITY |
|
LAPS | Lapse for failure to pay maintenance fees |
Free format text: PATENT EXPIRED FOR FAILURE TO PAY MAINTENANCE FEES (ORIGINAL EVENT CODE: EXP.); ENTITY STATUS OF PATENT OWNER: SMALL ENTITY |
|
STCH | Information on status: patent discontinuation |
Free format text: PATENT EXPIRED DUE TO NONPAYMENT OF MAINTENANCE FEES UNDER 37 CFR 1.362 |